Résumé
Remote sensing techniques in space applications use images that provide a large volume of information to collect data about the Earth. Hyperspectral images (HSIs) present a high volume of data and affect restrictions on the storage capacity and processing in space systems, highlighting the demand for compression. Systems that operate in space are susceptible to faults due to adverse conditions and require the implementation of protection techniques to mitigate these faults and ensure correct operation. This work implements a fault-tolerant CCSDS 123 HSI compressor using Triple Modular Redundancy (TMR) and Hamming Error Correcting Code (ECC) techniques to protect from Single Event Upsets (SEUs). We present the compressor in a standard version, followed by partially-protected and fully-protected versions, by applying hardening in different circuit components. We performed a reliability analysis through a fault injection campaign considering all implementations. The standard version showed 97.9% of executions with errors, which was reduced in the versions partially hardened, reaching no error propagation in the fully hardened version. In addition, all implementations accelerated the application, achieving a performance increase of up to 24× compared to the software solution.